What are some common methods used for nondestructive evaluation and why do you think x ray is better?
Figure 1. Digital image of a C-130 bracket showing crack detection using contrast enhancement (center) and edge detection (right).
Traditional nondestructive evaluation (NDE) methods include ultrasonic testing, eddy-current testing, visual inspection, magnetic particle testing, liquid penetrant testing, and x rays.
X ray happens to be one of the older technologies and also happens to be one that commands, at least in the U.S., about 40 percent of the total NDE instrumentation market. It is used for NDE in about eight major industry segments, including aerospace, automotive, and electronics.
Over the past several years, real-time or near real-time digital imaging techniques have been gaining importance in x-ray NDE (Figure 1). These include using CCD cameras as imaging devices for scintillators, which convert x rays to visible light. Then there is computed radiography, where a panel is exposed to x rays and that panel is read offline by an optical device in order to create the image.
Only in the past couple of years have large, flat-panel detectors based on amorphous silicon begun to be available (Figure 2). They offer several features that are important to many of the market segments. One is that it simulates a piece of film very nicely. The container isn't much thicker than an actual film holder and it can be placed in the same type of flat configuration as a piece of 8 X 10 film. People who are used to film will like the capabilities of the flat-panel array.
Figure 2. An 8 x 10-in. amorphous silicon imaging panel.
But, more importantly, detectors are continuing to improve in resolution. Pixel size is presently around 100 µm and will probably, in the next couple of years, go substantially below that. The size of the panel is also beginning to increase.
We're very positive that industry will accept this technology as it becomes more available over the next few years. Systems using flat-panel detectors will begin to see use in several key industries, such as the automotive, electronics packaging, and aerospace/aircraft industry.
Anytime you have a situation where the application requires robustness (i.e., the detector's ability to withstand shock, dirt, environmental changes, etc.), there is some hesitation whether you can use this technology. For example, this technology is not to the point where you can throw a flat-panel array into a ditch to make an x ray of a pipe, whereas a film holder could work in such an environment. But it will get there. The robustness is increasing all the time.
On the other hand, one thing it can withstand better than any type of film technique is radiation. It works well in a radioactive environment, such as inside a nuclear power plant, and because it's digital, you can do your background subtraction to get good images independent of the ambient radiation. It also lends itself to applications in radiation medicine where, for example, these flat-panel detectors can image a cancer location that is being treated by radiation. You can allow them to be exposed directly to the radiation beam without any significant degradation. You couldn't do that with a conventional CCD device.
Relative to other NDE techniques, I am not suggesting that you should use digital radiography to do something that, for example, ultrasonics can do well. Digital radiography is an excellent candidate for replacing many of the film-based radiography methods. If film radiography is a method that is being used routinely for either quality control or periodic examination, then it certainly makes sense to take a look at whether digital radiography, using flat-panel detectors, should be considered for that application.
One of the main reasons why flat-panel detectors are attractive is the larger dynamic range for intensities (4000:1) compared to film (100:1). You can utilize that dynamic range together with machine vision and pattern recognition to do automated detection of defects. We also want to tap into the huge wealth of image analysis technology that is out there and link it with the digital x-ray system so that automated decisions can be made, thereby improving throughput and enhancing productivity of the inspection process in general.
What kind of x-ray sources do you use?
We use a broad range of x-ray sources. For example, we have a system we've assembled recently in our laboratory for studying lower density materials. It has a 75 kV, microfocus x-ray tube, which makes it very good for certain types of plastics, tissues, and thin, low-density metals. X rays are formed from electrons leaving a heated filament that are then accelerated by the 75 kV potential difference, driving the electrons into the target. If an electron is stopped by the target with 100 percent conversion to photons, then the resulting radiation will have an energy of 75keV. This corresponds to a wavelength of about 0.2 Å, the smallest wavelength (largest energy) possible at this voltage. Because most accelerating electrons participate in less than 100 percent conversion, the mean wavelength of the resulting x rays is larger than this.
You can also go to the other extreme and use x-ray generators that are in the 300 to 10,000 kV range and do thicker or denser parts. There's no particular impediment to going to higher energies that can't be customized for that application. The thing we try to emphasize is the fact that, if you're putting the system together, you want your computer to control your source, to control your x-ray integration time, and also to essentially provide for any motion control you might be using for that particular application. But, there are no real impediments to covering a broad range of x-ray energies, say 20 kV to 10 MV.
It sounds like what you've got is an analog to a CCD detector on the back of the test object. Is that correct?
Yes, if you think of it as a large, thin CCD detector that will not degrade when it is exposed directly to x radiation. Right now, a CCD detector that is used for these applications is essentially a TV camera looking indirectly at a scintillator, which is being energized by the x rays. If you put the CCD detector directly in the x-ray beam, then it will degrade rather quickly over time, whereas an amorphous silicon panel will not.
Figure 3. The 12-bit sampling of the flat-panel detector gives a high dynamic range that allows many items of varying density to be imaged in a single exposure. Here we see an image of a steel resolution standard, the plastic envelope holding the standard, and even the masking tape holding the envelope to the surface of the flat panel.
With the amorphous silicon flat-panel array, you're able to place it directly in the beam and, secondly, it's quite large. You can tile four of these 8 X 10 devices together and make a very large, 40 X 32-in. flat-panel detector.
More importantly, as technology continues to evolve, there will be a reduction of the dependence on scintillators -- maybe even an elimination of them -- and this will come about by applying certain coatings to the amorphous silicon that will enable the direct detection of x rays. This would result in a reduction of the photons scattered to adjacent pixels, thus improving resolution. Also, as the "fill factor," which is the percentage of the actual pixel size that is sensitive to the x-ray, gets larger, the resolution will also increase. Right now, the technology we use gives resolution greater than 3 line pairs per millimeter (Figure 3). That's going to improve by at least two times over the next couple of years.
What is amorphous silicon compared to regular silicon?
The word "amorphous" simply means that it's not crystalline, which makes it quite different from the types of silicon that you're used to hearing about.
We do not manufacture the amorphous silicon. The amorphous silicon technology is essentially done by Xerox. Amorphous silicon is a semiconductor, like crystalline silicon, but it is much less susceptible to degradation from exposure to radiation. The amorphous silicon serves as a platform for thin film transistors, which are the key components of each light-sensing pixel. It is also possible to make amorphous silicon in larger sizes than crystalline silicon wafers, hence large flat-panel arrays enable large detectors.
Can you take us through this detection process step by step?
Typically, you have a lead-lined enclosure that provides shielding from the x rays incident on the test part. You place your part in the enclosure, between the x-ray generator and the flat panel (Figure 4). The part can be directly in contact with the flat panel, which is like putting a piece of film behind it, or you can move the part away from the panel, which gives you additional magnification. Once you turn your x-ray generator on and expose your part for a preselected amount of time, the x-ray image from the flat-panel detector will immediately appear on the screen of your computer.
Figure 4. Typical configuration of flat-panel imager that creates geometric magnification by using a microfocus x-ray source and separating the imager from the location of the test part.
If it's one of many identical parts that are going to be x rayed -- in other words, if it's a quality control type of application -- then you can train your system to essentially memorize the acceptable variations in density of that part. Any variation that is outside a specified tolerance will then be automatically recognized by the computer and you'll be alerted that there is an anomaly in that part. This, of course, enables you to find holes and cracks that aren't supposed to be there. With a certain amount of motion control, certain disbonds can be detected. Most importantly, you can eliminate the need for having someone view and make a decision on every x ray. The computer can catch something that is different and then a person can intervene. So, this technology has the capability of substantially improving the throughput and reducing cost (Figure 5).
Figure 5. TPL's Digital X-ray Imaging (DXI) systems provide a single graphical user interface for source control, image acquisition, motion control, and image analysis.
What's a sample exposure time? Is it seconds or minutes?
We're talking seconds here. For example, we recently did some ammunition fuses that consisted of a variety of metal and plastic parts. Those fuses were exposed for about 5 seconds at around 60 kV. That was plenty of energy to illuminate all the things that needed to be seen on the computer and then make a decision as to whether the fuse was properly assembled and, most importantly, not armed.
Because this is a digital image, you can store it on computer and easily transmit it all over the place?
That's correct. You get the image immediately. It's on the screen immediately. It's in the computer immediately. If you want to send it to somebody else, you can send it so that they can have it.
Obviously, one of the nice things about digital radiography is that you don't need to have huge rooms for archiving your film. Also, there's a substantial reduction in waste since you're not developing film.
Together with Xerox-Palo Alto Research Center (Palo Alto, CA) and ThermoTrex Corp. (San Diego, CA), TPL is a participant in a $13 million, four-year project sponsored by the National Institute of Standards and Technology (NIST). The project will advance the capabilities of the flat-panel detectors and is studying ways of applying this technology to various industry segments. TPL is working with collaborators who have specific applications in NDE and who would like to adapt the digital x-ray techniques to their applications. We look at their NDE needs and suggest system features, such as the type and size of flat-panel detector, pattern recognition software, and motion control.
In that scenario, there have to be standards on digital radiography that facilitate the transition from film-based to filmless technology.
We continue to look for potential collaborators for NDE applications in several industry segments: electronics packaging, electric power production, aerospace, chemical, and petrochemical. We'd also like to add metal inspection industries that have production line types of requirements for automated inspection.
Our business in this area is twofold. We're involved in this consortium to basically further the applications of this technology. We're also providing integrated systems consisting of the sources, the x-ray detection panels, the software, the computers, motion control; and we tailor the software package to the application. This technology will help industries see defects easier, establish better quality control, and facilitate the storage and transmission of images.
William Hartman is vice president of the Advanced Technologies Division of TPL, Inc. (Albuquerque, NM). He was interviewed by Frederick Su.